News · legal — By WATMM Desk — · Updated

'Dark Horse': how a $2.8M verdict against Katy Perry was wiped out over an unoriginal ostinato

The Ninth Circuit has upheld the decision to throw out a jury's $2.8 million copyright verdict against Katy Perry, holding that the eight-note figure 'Dark Horse' shares with Flame's 'Joyful Noise' is built entirely from commonplace musical elements. The opinion is the clearest plain-English account yet of how courts separate a song's building blocks from what a songwriter can own.

Key facts

  • What happened: on 10 March 2022 the Ninth Circuit affirmed the March 2020 order that vacated a 2019 jury award of about $2.8 million and entered judgment for Perry and her co-defendants.
  • How: under the extrinsic test the court stripped out unprotectable elements (the minor scale, an even eight-note rhythm, synth timbre, a 3-3-3-3-2-2 pitch run) and found nothing original left in how they were combined.
  • Why it matters: judges, not juries, decide what part of a song is protectable, so a jury's view that two songs sound alike can't carry a claim built on musical building blocks.

A jury decided in 2019 that Katy Perry's "Dark Horse" copied a Christian hip-hop song and awarded about $2.8 million. On 10 March 2022 the Ninth Circuit confirmed that the award is gone for good. In Gray v. Hudson, a three-judge panel (Clifton, Milan D. Smith Jr. and Watford, with Smith writing) affirmed the trial judge's decision to set the verdict aside and enter judgment for Perry and her co-defendants. The short version: the repeating eight-note figure the two songs share is made entirely of material nobody can own.

The opinion is worth reading even if you never go near a courtroom. It's the clearest explanation we've seen of how US courts decide which parts of a song are protected, and it leaves the door open a crack for combinations of common elements.

What happened, in order?

The facts below come from the opinion unless another source is linked.

  • 2007–08. Chike Ojukwu recorded a simple tune on a free music website and sold it to Marcus Gray, who performs as Flame. Gray used it as the ostinato, a repeating instrumental figure, in "Joyful Noise", released in 2008 on Our World Redeemed. The song was no radio hit but drew millions of views on YouTube and MySpace, and the album was nominated for a Grammy in 2009.
  • 2013. Perry and her collaborators made "Dark Horse". It became a hit and led to a music video and her 2015 Super Bowl halftime show.
  • 2019. After a bifurcated jury trial from 17 July to 1 August, the jury found access and substantial similarity and awarded the plaintiffs 22.5% of the defendants' net profits from "Dark Horse", about $2.8 million in total.
  • 17 March 2020. Judge Christina A. Snyder of the Central District of California vacated the verdict and granted judgment as a matter of law (JMOL) to the defendants, as Music Business Worldwide reported. She also conditionally granted a new trial in case the JMOL was reversed. Her 32-page decision rejected every challenge to the verdict except one: the ostinatos weren't substantially similar in any protected way.
  • 10 March 2022. The Ninth Circuit, reviewing the question afresh (de novo), affirmed. It said it didn't need to decide anything else, including whether Perry's team ever had access to "Joyful Noise".

What do the two ostinatos actually share?

Both are in a minor key, though different keys. In scale degrees, the "Dark Horse" figure runs 3-3-3-3-2-2-1-5. "Joyful Noise" alternates 3-3-3-3-2-2-2-1 and 3-3-3-3-2-2-2-6. So the overlap is the opening 3-3-3-3-2-2, and both are played in a completely even rhythm, every note the same length. The defence musicologist, Lawrence Ferrara, pointed out that "Jolly Old Saint Nicholas" and "Merrily We Roll Along" use the same 3-3-3-3-2-2 run in a major key. That evidence went unrebutted at trial.

How does the extrinsic test work, in plain English?

In the Ninth Circuit, a plaintiff without direct evidence of copying has to show access and "substantial similarity". Similarity has two halves, and both must be met:

  1. Extrinsic: take both works apart, set aside everything copyright doesn't protect, and compare what's left using objective criteria. This is a question judges can and often do decide as a matter of law.
  2. Intrinsic: would an ordinary listener, without expert help, find the expression similar? This is the jury's territory, and appeal courts rarely second-guess it.

The plaintiffs' case failed at step one. Their own expert, Todd Decker, listed the shared features as the eight-note length, the rhythm, the scale degrees, the melodic shape, the timbre and the ostinato's place in the mix, and conceded that no single one of them would have been enough. The court then went through them:

  • The minor scale is a public building block.
  • Eight notes in an even rhythm is a "trite" choice. Decker himself called the rhythm a "relatively simple rhythmic choice" that "no composer's entitled to monopolize."
  • Texture and placement in the mix are too abstract to protect.
  • Timbre belongs to the performance or recording, not the composition. Copyright in a song, the court wrote, doesn't let anyone own the sound of a synthesizer any more than a trombone or a banjo.
  • The pitch sequence on its own is not a melody, since a melody needs rhythmic organisation too. The court leaned on Copyright Office guidance that short phrases of only a few notes aren't copyrightable even if novel, and noted that chord progressions aren't protected either.
  • Melodic shape, tension on 3 resolving through 2 towards 1, reflects common rules of consonance, the musical equivalent of a film trope.

Finally, the court asked whether the combination was original. It can be: the opinion repeats circuit precedent that even seven notes isn't too short as a matter of law. But here the shared portion was "nothing more than a two-note snippet of a descending minor scale, with some notes repeated" in a flat rhythm. Protecting it, the panel said, would amount to a monopoly over two-note pitch sequences or even the minor scale, given how few choices an eight-note repeating figure allows. A footnote is careful to say the court isn't suggesting "Joyful Noise" as a whole lacks original expression.

The panel also brushed off two procedural complaints: that the trial judge relied on amici's post-trial database searches (a single "see also" citation that changed nothing), and that she couldn't grant JMOL after denying summary judgment on the same issue (no authority says so).

What we think

This is the right result, and it should make songwriters safer, not more exposed. The headline lesson, as the National Law Review and Stanford's case summary both put it, is that commonplace elements and unoriginal combinations of them can't carry an infringement claim. The bigger point is who that protects. If a repeated 3-to-2 step in even notes were ownable, every producer working in a minor key would be one plausible access theory away from a jury trial. We'd bet most working musicians are likelier to be accused than to accuse, so a high bar for building-block claims protects them more often than it hurts them.

But notice who carried the cost of getting there. Flame and his co-writers won a jury verdict, then lost it about seven months later, then waited two more years for the appeal. The opinion makes the trial judge the gatekeeper of what's protectable, which is correct in principle, yet it means a lay jury's sense that two songs sound alike can be worth nothing years later. For independent artists, the practical advice is to have the extrinsic question tested hard, ideally by a musicologist briefed on copyright law and not just on music, before spending years in litigation. Decker's candid answer that no single element sufficed was, as the court put it, "arguably sufficient" on its own to sink the case.

The timbre point matters most for producers. The court drew a firm line between the composition and the recording. If what makes two tracks feel alike is the synth patch, the register and where it sits in the mix, that's a sound-recording question, and claims about the composition won't reach it. A high, "pingy" synth sound, to use the expert's word, isn't something a songwriter can claim; copying an actual recording is a different matter under a separate copyright.

What's still open is the combination. The panel didn't say short figures are never protectable. It said this one wasn't. A riff with a distinctive rhythm, an unexpected interval and a recognisable shape could still qualify. The safest reading is that originality lives in the specific choices that the tradition doesn't hand you, and that's where to put your effort if you want something you can defend.

Sources

Sources

  1. Gray v. Hudson, No. 20-55401 (opinion), U.S. Court of Appeals for the Ninth Circuit (primary source). Accessed 23 Oct 2026.
  2. Gray v. Hudson, Stanford Libraries, Copyright and Fair Use. Accessed 23 Oct 2026.
  3. So You Wanna Play Copyright? Joyful Noise Ostinato Isn't Original Expression, The National Law Review. Accessed 23 Oct 2026.
  4. Katy Perry's $2.8m Dark Horse copyright infringement verdict has been overturned, Music Business Worldwide. Accessed 23 Oct 2026.

Researched with AI assistance, written and checked by WATMM Desk.

Change log

  • : Backfilled in October 2026; later sources were added for context.

Subscribe to We are the music makers

One email a week: the music-tech and AI-in-music news that mattered, and what we think about it.

Sign-ups open soon.